Robert Vaser

Hi Tom, I run your data on Ubuntu 16.04. 10 times in a row, it worked in all cases. No idea what is causing the problem on RHEL 7.6 :/...

Hi Peter, the memory is bound by the cumulative size of all three files needed as input plus some epsilon. For your case, I would say that 500Gb of RAM...

Hi Ka Ming, thanks for reporting this. How long was the CIGAR string? The parser expects the whole line to be shorter than 3000 characters. Best regards, Robert

Hi, try adding cuda path when running cmake, something like this `-DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-10.0/`. Best regards, Robert

Hello, I would advise you to split your read set into small subsets depending on your RAM size (e.g. 5GB) and map the whole read set on all of the...

Hello Anne, splitting your files won't decrease the runtime, only the memory consumption. The log contains "thread_id = 0" because the first thread prints the overall progress of the program...

Well, error correction is slower than consensus/polishing as it uses all possible overlaps between reads. If your sequenced data is from a repetitive genome and you are using minimap for...

@wenjiexiaomeng, racon `1.2.0` has a wrapper script to enable easier correction for large files (you can split target sequences to decrease memory consumption). Also, all racon versions beyond `1.0.0` should...

Hello Adam, obtain pairwise overlaps with minimap2 by using `minimap2 -t num_threads -ax ava-pb/ont --dual=yes reads reads > overlaps.sam` and run `racon -t num_hreads -f reads overlaps.sam reads > polished_reads.fasta`....

The polished sequences size will change, but it won't extend sequence ends with other sequences that have an prefix-suffix overlap or vice versa (if I understood correctly).